The purpose of this study was to compare the ear-based rectal temperature measured with a tympanic thermometer with the rectal temperature measured with a glass mercury thermometer in order to test the accuracy of tympanic thermometer and to determine relationship among rectal, axilla, and abdominal temperature in neonates. The samples consisted of thirty four neonates admitted to the neonatal intensive care unit and nursery at an university affiliated hospital. The mean age of the subjects was 4.9 days. The ear-based rectal temperatures were taken with a tympanic thermometer in rectal mode (First Temp Genius 3000). Rectal and axilla temperatures were taken with a glass mercury thermometer, Abdominal temperature was continuously monitored with the probe connected to the servo controller of incubator. The results of the study can be summarized as follows : 1. Intrarater comparison : Agreement between the first and the second ear-based rectal temperature was 97% within 0.1$^{circ}C$. 2. Comparison of ear-based rectal temperature and the rectal temperature from a glass mercury thermometer : ear-based rectal temperature ranged from 36.95$^{circ}C$d to 37.95$^{circ}C$, with a mean of 37.58$^{circ}C$(SD＝0.22$^{circ}C$). Rectal temperature from a glass mercury thermometer ranged from 36.2$0^{circ}C$ to 37.2$0^{circ}C$, with a mean 36.75$^{circ}C$(SD＝0.29). The mean difference between both temperatures was 0.84$^{circ}C$. The correlation coefficient between both temperatures was r＝0.77(p＝0.00). 3. Comparison of rectal and axilla temperature : Axilla temperature ranged from 35.8$0^{circ}C$ to 37.1$0^{circ}C$, with a mean of 36.55$^{circ}C$. The mean absolute difference between the rectal and axilla temperature was 0.23$^{circ}C$. The correlation coefficient between rectal and axilla was r＝0.67. 4. Comparison of axilla and abdominal temperature : Abdominal temperature ranged from 36.2$0^{circ}C$ to 37.0$0^{circ}C$, with a mean of 36.58$^{circ}C$. The mean absolute difference between axilla and abdominal temperature was only -0.03$^{circ}C$. Findings of this study suggest that ear-based rectal temperature overestimates the actual rectal temperatures in neonates. Therefore, the interchangeble use of both temperatures in clinics seems problematic. The site offset(adjustment value) programmed in rectal mode of the tympanic thermometer needs to be readjusted. Choosing one optimal site for temperature measurement for each patient, and using the specific site consistently would result in more consistent measurements of changes in body temperature, and thus can be more effective in diagnosing fever or hypothermia.